[SPEC] Backwards-compatible metadata in Gemini

[avr at geminet.org]

On Fri, Feb 26, 2021 at 11:44:38AM +0100, Stephane Bortzmeyer wrote:
> On Wed, Feb 24, 2021 at 10:11:07PM +0200,
>  Lars Nood?n <lars.nooden at gmx.com> wrote 
>  a message of 27 lines which said:
> 
> > The actual contents of the key-value pairs, the actual metadata itself,
> > would not be part of the spec, not even the keys.
> 
> May be not in "The Official Spec" but it certainly needs to be
> somewhere, because otherwise people would tag the same information
> differently ("last-modified", "last-change", "updated"), then
> destroying its utility. 

I'd say it is diminishing its utility, and even then mostly for third parties (like indexers).
I also believe that useful metadata will roughly converge to a format within a
natural language, like we're already seeing with meta-data like "Tags:". Sure, there are variations,
but it's not unreasonable to say these are the responsibility of the third party to pick up on. 

And if some writer really wants to be indexed by a particular engine, that engine
should put up a list of useful metadata for writers to use. Different engines can
copy each other, or be more distinctive in their meta-data requests.

Now e.g. chinese writers can freely use "??:" for tags, meaning they probably don't 
care about being indexed/archived by non-chinese engines, but if an all-languages
engine still want to index the page, they can if they put in the work.[0]

So,

What are the benefits of a consolidated meta-data format for the two most important parties
within Gemini: the writer and the reader?

Why are the third parties (indexers, archivers, etc) for whom meta-data could be
important not working on this by themselves? They can write and publish rules, and
then try to get the writers onboard. And writers can decide for themselves if the
extra trouble is worth the extra discoverability/better archiving, etc, forcing third 
parties to co-operate and to keep it simple.

In short, I don't understand the need for regulation when non-regulated convergence
seems a natural outcome that will fit the various use cases better through an evolutioanry
process. Granted,  waiting for convergence is a lot slower, but does faster adoption  
really matter with Gemini?

[0]"??" is a dictionary translation for illustration purposes.